1. Field of the Invention
The present invention relates to video recognition technology, and more particularly, to a method for providing a multimedia stream.
2. Background of the Related Art
As the digital video technology and the image or video recognition are developed, a user can access a desired information by searching or browsing certain particular segments of the whole stream. In other words, the user can effectively understand the contents of a predetermined video stream by reproducing the video stream and browsing only a desired portion of the video stream or indexing a digest of the video stream without viewing the whole video stream.
For this purpose, processing a multimedia stream in order to describe a content-based data with the whole multimedia stream or its segments and search the content-based data when the user requests a desired multimedia stream has been studied. The content-based data includes an information on the entire contents of a program, such as information on a production company, director and cast, and a semantic information on each segment in a multimedia stream. Such semantic information can include information on the appearance/disappearance of an object or background, occurrence of an event, and relation between objects and events, throughout the multimedia stream.
When the above information is stored in a content-based data region for a multimedia stream, the user can selectively search a desired stream, or browse only a desired portion in a multimedia stream, by retrieving the content-based data. Therefore, if there is a user's request, saying that, for example, “show only the scene in which a main actor appears in close-up,” an apparatus employing a method for providing a multimedia stream based on the content-based data outputs the corresponding segment. Here, the content-based data contains an appearance of an audiovisual object, status of an audiovisual object, event, background, and segment information. In addition, the segment information contains start and finish points of time, audiovisual objects representing a segment, text description for a segment, and events shown in a segment.
In U.S. Pat. No. 5,913,013, a content map is constructed by using a method of setting a level of a violence scene, nude scene, portion in which an actor and/or actress appears, etc. and setting a code of segments, and the user can easily browse only the desired portion.
The method for providing a multimedia stream according to the related art will be described in more detail.
FIG. 1 is a diagram illustrating an example of a method for describing a content-based data for a multimedia stream according to the related art, in which a number of comments (comment 1, comment 2, . . . ) are described along with a segment information. As shown in FIG. 1, the comment means an appearance of an audiovisual object, status of an audiovisual object, event, background, and segment information, and Si (i=0, 1, 2, . . . ) means a segment. However, while the above content-based comment helps the user to easily search a desired portion, it is difficult for the comment to properly meet the user's need for understanding the whole contents of the multimedia stream within a short time.
To solve the above problem, an algorithm for describing a multimedia stream in a multi-level digest segment information scheme is proposed.
FIG. 2 is a diagram illustrating a related art multi-level digest segment information scheme and the corresponding multi-level digest stream, in which digest levels (0, 1, 2, . . . ) according to the level importance of segments of a multimedia stream and multi-level digest streams (H0, H1, H2, . . . ) are described. As shown in FIG. 2, the digest stream (H0) consists of digest segments of the digest level (0), the digest stream (H) consists of segments of the digest level (0, 1), and the digest stream (H2) consists of segments of the digest level (0, 1, 2).
The multi-level digest segment information scheme shows a digest stream in multi-levels by giving an importance level to each digest segment. That is, the multi-level digest segment information scheme describes each digest segment not by using a physical data with respect to multi-level digest segments, but by using a segment information and importance level information of the digest segments. However, these related art methods are inconvenient in that the user has to input a filtering condition or search condition on every filtering or search operation, using a content-based data described in a content-based data region or a content map, in order to browse a user-desired segment or contents.
To mitigate the inconvenience, another related art system employs a method for automatically recommending a user-preferred program or providing only segments which the user is likely to want by combining user profile information based on the history of the user, a content map for a multimedia stream, etc. The user profile is stored in a portable nonvolatile memory device such as smart card, a fixed non-volatile memory of a user's terminal, or a nonvolatile memory space of a multimedia server. In addition, the user profile is automatically learned from the browsing history of the user, or is updated by the editing of a pre-registered user.
However, most users want to save time by browsing a digest of a multimedia stream before browsing the whole multimedia stream, in order to determine whether to view the whole multimedia stream. With a multi-level digest stream information scheme, content providers can describe multiple versions of highlights with small amount of additional data and the users can request a variety of digest streams of the quantity of 10, 20, and 30 minutes, with respect to an original multimedia stream with a running time of two hours. However, since the method for providing a digest stream gives a level to segments in the order of importance in understanding the whole contents irrespective of various inclinations of the users, a digest stream reflecting each user's preference or character cannot be provided.
Meanwhile, when both of the multi-level digest stream information and the content-based data for the multimedia stream are used, it is possible to meet a request for a multimedia stream based on a user query. However, the related art method for providing a multimedia stream require additional actions, take longer to process and are inconvenient in that the user has to input a condition of a digest, whenever necessary, using a natural language or other interface.
The above references are incorporated by reference herein where appropriate for appropriate teachings of additional or alternative details, features and/or technical background.